Adaptive Acoustic Event Location System Of Wind Turbine Blade Under Complex Background Noise

The blade is the core component of the wind turbine to obtain wind energy.Due to frequently affected composite alternating stress,it is easy to surface sand erosion,girder deformation,trailing edge cracking.With the increase of the wind turbine service time,accidents caused by blade failures have occasionally occurred.The lack of maintenance and detection is the main cause of malignant accidents in the industry.Compared with the contact detection methods such as vibration,fiber Bragg grating and ultrasonic detection,the non-contact acoustic detection method based on microphone has the advantages of flexible sensor installation,high sensitivity and good economic applicability.Owing to the influence of occasional noise such as strong wind noise,yaw,pitch and changes in wind turbine operating conditions,the pretreatment method for blade acoustic diagnosis is difficult to capture the dynamic characteristics of the blade acoustic events,which causes deviations in the location of wind turbine blade acoustic event.Accordingly,this paper presents an adaptive location system for wind turbine blade.The specific research contents and results are as follows:(1)Aiming at the issue of complex background noise interference in the acoustic signal of wind turbine blade,a noise reduction method of wind turbine blade acoustic signal based on adaptive maximum correlation kurtosis deconvolution is proposed.Introducing the idea of maximum correlation kurtosis deconvolution difference filtering,an automatic cycle detection algorithm based on the correlation entropy index of envelope spectrum is designed,and the cycle parameters are estimated in real time by using the iterative characteristics of deconvolution.Combining multiscale sample entropy to construct evaluation criteria for the termination condition of deconvolution iteration,an improved maximum correlation kurtosis deconvolution algorithm is established.The experimental results show that compared with adaptive spectral subtraction,high-pass filtering and maximum correlated kurtosis deconvolution,the algorithm has a significant improvement in the noise reduction effect of acoustic signal of wind turbine blade.(2)An adaptive positioning method based on multi-source information fusion and dynamic threshold is proposed to solve the problem of wind turbine blade acoustic events location and extraction in variable working conditions.The Relief algorithm is introduced to obtain the feature parameters of multi-source feature fusion,which solves the problem that the single feature parameter has limited ability to describe the difference between the target signal segment and the background segment.An adaptive dynamic threshold positioning strategy is established by combining Wasserstein distance and fuzzy C-means clustering.The experimental results show that compared with the traditional threshold method,fuzzy C-means clustering threshold method and spectrum-based centroid location method,this method has.higher reliability for the location of wind turbine blade acoustic events.(3)Combined with the actual application requirements,the overall architecture of the intelligent analysis platform for blade acoustic events is constructed by adopting the modular idea.By transplanting the front-end preprocessing and positioning algorithm,the application software for the design pattern of front-back end separation and responsive layout is established.Based on the measured data collected by the wind farm,the test verifies the functional correctness of the core modules such as noise reduction analysis,location extraction,audio playback,and setting management.